The later immigrants arrive in North America and begin to learn English as a second language (L2), the less accurately they produce and perceive certain English vowels and consonants, the stronger their overall degree of foreign accent tends to be, and the more difficulty they may have recognizing English words. Such effects are usually attributed to the passing of a critical period. This research will evaluate an alternative account, that of the Speech Learning Model (SLM). The SLM proposes that age- related effects on L2 speech arise from the development of the L1 phonetic system, not a diminished ability to learn speech. More specifically, the SLM proposes that as L1 phonetic categories are refined, they become more likely to assimilate L2 sounds and thus to block formation of the phonetic categories needed for the accurate production and perception of speech sounds. The first proposed project will examine native Spanish adults who recently immigrated to the U.S. at 6-month intervals over 7 years. Among the hypotheses to be tested are that (1) Overall degree of foreign accent will decrease as the ability to produce particular English sounds improves; (2) Improvements in segmental production will be observed only if the ability to perceive certain English sounds has also improved; (3) Improvements in the perception of English sounds will be accompanied, or preceded by the detection of phonetic differences between sounds found in English and Spanish. Another project will examine 240 native Italian adults who are all highly experienced in English but who differ according to their age of first exposure to English (range: 10 to 30 years). The SLM predicts that the native Italian participants' degree of foreign accent will increase linearly to the end of adolescence, then level out. The model also predicts that a significant correlation will exist between degree of foreign accent and the scores obtained on two perceptual tests. The final project will examine four groups of native Italian speakers who differ according to their age of first exposure to English and self-reported native language (L1) use. Eight experiments will test for the formation of phonetic categories by examining the production and perception of English vowels, and the recognition of English words. By hypothesis, early bilinguals will outperform late bilinguals, and individuals who now use their L1 seldom will outperform those who continue to use their L1 frequently.